Download Spanish Light Verb Constructions: co-predication with

Spanish Light Verb Constructions: co-predication with syntactically
formed complex predicates
Rodrigo Romero Méndez
University at Buffalo,
The State University of New York
Rodrigo Romero-Méndez
Table of contents
1 Introduction...................................................................................................................... 1
1.1 Previous studies ........................................................................................................ 2
1.2 The proposal.............................................................................................................. 5
2 Semantic and syntactic properties of LVCs..................................................................... 6
2.1 Two main types of LVCs.......................................................................................... 6
2.2 Light Verb Constructions are not lexical phrases. .................................................... 7
2.3 Number of arguments in LVCs............................................................................... 10
2.4 LVCs and their syntactic status............................................................................... 14
2.5 LVCs, auxiliaries and Clause Union....................................................................... 16
2.6 Light verbs and Collocations .................................................................................. 20
2.7 Some regularity: families of LVCs ......................................................................... 22
2.8 Summary ................................................................................................................. 25
3 Light verb constructions in Role and Reference Grammar ........................................... 25
3.1 Semantic Decomposition in RRG........................................................................... 25
3.2 Initial analysis of Light Verbs................................................................................. 28
3.3 The semantics of predicative nouns........................................................................ 31
3.4 Verbal schematism and qualia in the formation of LVCs....................................... 34
3.5 Semicompositionality and semiproductivity in LCVs............................................ 37
3.6 Light verb interpretation vs. heavy interpretation................................................... 40
4 Concluding remarks ....................................................................................................... 43
5 References...................................................................................................................... 44
2
Presentation
The present paper deals with combinations of verbs plus nouns known as
Light Verb Constructions (LVC), such as Bill took a shower, in which the
interpretation of the eventuality depends on both the noun and the verb.
Following Butt & Geuder (2001), I will argue that LVCs are syntactically
formed complex predicates. In LVCs, the event denoting noun combines
its LS with the LS from the verb, creating an LS that describes a single
eventuality. In this way, the argument structure for the LVC is not directly
and solely licensed by the noun, but rather it is yielded by both syntactic
components. I will argue that in these constructions are semi-productive
and semi-compositional, which is better captured using a system of lexical
decomposition, in this case Role and Reference Grammar (Van Valin and
LaPolla 1997, Van Valin 2005).
1 Introduction
In this paper I will discuss the main characteristics of Light Verb Constructions (LVC),
cases like Pat made new comments on the case, in which the noun is combined with a
verb in order to produce a syntactically complex predicate. In other words, while the
semantic content for the eventuality comes from both the noun and the verb, the noun is
still the head of a noun phrase and it still preserves the characteristics of regular noun
phrases in object position.
The term light verb construction is rather unfortunate for several reasons. First of all,
it suggests that the verb has no meaning, which is not true, as argued in this paper.
Second, it has been applied to many phenomena that, albeit related, are different form
language to language. While the term has been used for verb+verb (V+V) constructions
in some languages (Mohanan 1997, Butt & Geuder 2001, Pensalfini 2004, inter alia), it is
Rodrigo Romero-Méndez
applied to only verb+noun constructions other languages. Sometimes linked with the
previous problem, verbs appearing in an LVC, or light verbs (LV), have been conflated
with auxiliary verbs.
Here, I will argue that in Spanish LVCs are restricted to V+N constructions. In
addition, LVs are different from auxiliary verbs for several reasons, but they are not a
special subclass of verbs. Rather, an LV is such insofar as it is part of an LVC.
1.1 Previous studies
The notion of light verbs goes back to Jespersen (1964:117), who observed that some
deverbal nouns are combined with verbs that are light in their semantic content, such as
have, to form a complex predicate. Even tough more than forty years have passed, it is
still believed that in LVCs the noun is the main predicate and that light verbs have no
semantic content, two notions that are not accurate, as explained in the next section.
There is no single definition of LVCs and in general there is no agreement as to what
counts as an LVC.
Most of the previous analysis on LVCs has been centered on some syntactic
properties of LVCs, especially on how the semantic arguments from the noun are
syntactically licensed by the verb. For many scholars, such as Cattell (1984) and
Grimshaw and Mester (1988), the verb in an LVC is only a functional element which
allows the noun to be the main predicate and to license its arguments, something
otherwise impossible. This comes with the idea that those verbs have no meaning. Not
always differentiated from the previous position, from different frameworks other studies
have proposed that in an LVC should be considered along auxiliary verbs (Alba-Salas
2001, Alonso Ramos 2004).
2
Spanish Light Verb Constructions
However, it was also observed that the matrix verb imposes some selectional
restrictions. In this respect, Grimshaw and Mester (1988:229) pointed out that
the influence of the [light] Verb itself is detectable in subtle meaning
changes. For example, although a spider can walk, a spider does not
normally take a walk. This difference presumably reflects the influence of
the argument structure of take on the interpretation of the complex.
It has been argued that there are other cases in which some of the arguments of the LVC
are not licensed by the noun alone, but rather by the verb. Then, in addition to imposing
selectional restrictions, some verbs also have a partially specified argument structure.
Constructions of this type were called semi-light verb constructions (Di Sciullo and
Rosen 1990, Pelletier 1990).
Two mechanisms have been proposed for combining the arguments for the noun into
the verbal argument structure. For Grimshaw and Mester (1988) this is done via
argument transference, according to which the noun transfers its θ-roles to the verb. This
operation follows from the fact that the light verb is thematically incomplete, and the
noun alone cannot project its arguments beyond its phrasal level; so the θ-roles from the
noun have to be transferred to the verb. The second mechanism is argument substitution.
According to Di Sciullo & Rosen (1990:110) “A light verb is one that has only
unspecified variables at Argument Structure”. Consequently, a light verb construction is
the result of the substitution of one of the unspecified variables of the light verb with the
argument structure of the second predicate. A solution along these lines is offered in (1),
where u represents the unspecified arguments and the sub-index represents a control
relation. There have been other proposals along the same lines (Rosen 1990, Mohanan
1997, Samek-Lodovici 2003).
3
Rodrigo Romero-Méndez
1. V1
V2
(ui, u)
( xi, y) → [V1 V2] (x, y)
Butt & Geuder (2001) consider LVCs as complex predicates, where LVs are semi-lexical
heads. Contrary to most proposals, they define LVs as a class of verbs in Urdu, different
from auxiliaries. LVs are a type of co-head, where both elements jointly determine the
“predicative power of the lexical predicate” (Butt & Geuder 2001:333). The notion of cohead is similar to co-subordination in RRG (Van Valin and LaPolla 1997) or to colexicalized compounds for Givón (2001).
Specifically for Spanish, LVCs are regarded as unanalyzable idioms for most
scholars (see for example Piera & Varela, 1999; Gómez Tórrego 1999). There are,
however, two important studies: a comparative study of LVCs in Romance languages by
Alba Salas (2001), and another devoted only to Spanish by Alonso Ramos (2004).1
Alaba-Salas (2001) proposes within Relational Grammar that in all LVCs the noun
should have a deep subject, and the “argument transference” is achieved by demotion.2
There are two fundamental problems: first, not all nouns in LVCs have a deep subject;
second, he has to postulate that action nouns in LVCs are in fact verbs, for which there is
no independent motivation. Alonso-Ramos (2004) proposes rather that LVCs are semiphrases in which the meaning of the verb is already contained in the meaning of the noun.
The downside is that ultimately LVCs are treated as lexical phrases.
Finally, it is worth pointing out that the notion of light verb has been extensively
used in recent years in the Minimalist Program (Chomsky 1995, Bosque 2001, Adger
1
In essence, Alonso Ramos’ (2004) book is a modified version of her Ph.D. dissertation (Alonso Ramos
1998).
2
Alba-Salas (2002) does not use the term “deep subject” but rather he refers to them as a 1-relation within
the P-sector of the noun. He does not use the term “argument transference” either, although rearrangements
in the argument structure do the same job.
4
Spanish Light Verb Constructions
2003). However, such a notion corresponds to an empty node in the tree (the “little v”)
and has in fact nothing to do with light verbs in the present discussion.3
1.2 The proposal
I will consider light verb constructions as syntactically formed complex predicates in
terms of a merged logical structure representation. This idea has been suggested already
by Butt & Geuder (2001), though LVCs should be restricted to V+N constructions in
Spanish. I will argue that both the noun and the verb are co-predicators since both of
them contribute to the LS (cf. Van Valin 2004). However, in syntactic terms, only the
verb is the head of the VP while the noun is a verbal argument (in most cases the direct
object). In other words, in syntactic terms LVCs have a simple nucleus, with only one
lexical item as the head, being the verb, in opposition to cases of co-subordination that
might have a complex nucleus or a complex core (i.e. formed by two verbs). Related to
this point, I will argue that syntactically LVCs are different from auxiliary verbs
constructions in Spanish.
In terms of the LVC formation, the verb takes as its argument an event or state
denoting noun, which will be called the predicative noun (or P-noun). This P-noun
combines its meaning with part of the meaning provided by the verb. In other words, the
Logical Structure (LS) is the result of the co-composition (Van Valin 2004) of the LS for
the verb with the LS for the noun. In addition, the argument structure from the resulting
LVC is not directly licensed by either the noun or the verb alone; rather it follows from
the LS.
3
It is necessary to say, though, that the motivation for saying that there is a vP level comes from the
analysis of VP-Shells and causatives. For some authors causative verbs also count as light verbs.
5
Rodrigo Romero-Méndez
As will be explained, there are more patterns in LVCs than what has been previously
assumed. So, it is possible to identify some “families” of LVCs. Even though there are
regularities inside those families, and even productivity, there is a great deal of
idiosyncrasy across groups. Thus, the combination between the verbal and the nominal
LS is not completely unsystematic but it is not completely compositional either. Rather,
they are semi-productive and semi-compositional. Finally, in order to argue for complex
predicate formation and capture the characteristics of Spanish LVC, it is necessary to
have a theory that allows semantic decomposition.
Finally, two terminological notes are in order. In many studies the term heavy verb is
used in opposition to light verbs. In this study I will also use it for the sake of contrast,
and it does not have any theoretical status. Second, even though the categories of
“subject”, “direct object” and “indirect object” are not technically defined in RRG, I will
use them in order to keep things simple.
2 Semantic and syntactic properties of LVCs
2.1 Two main types of LVCs
The first thing to be noted is that there are two main classes of LVCs, depending on the
type of predicative noun (P-noun) that the verb takes as its argument: in some cases it is a
state denoting noun (such as miedo ‘fear’) while in other cases it is an event denoting
noun (such as patada ‘kick’). This first semantic classification has, as one might expect,
some syntactic correlations. Even though in both cases it is possible to modify the noun,
they present different syntactic characteristics, not highlighted in previous studies (AlbaSalas 2002, Alonso Ramos 2004). Not surprisingly, an LVC having an event denoting
noun might alternate between a cumulative interpretation with a plural noun (2a) and a
6
Spanish Light Verb Constructions
quantized interpretation a numeral (2b) or a quantifier (cf. Krifka 1992). In addition,
event denoting nouns might also refer to specific events and thus they also accept the
definite article (2c).
2. a. Sergio les da besos a las muchachas.
‘Sergio gave kisses to the girls’
b. Sergio les dio tres besos a la muchacha.
‘Sergio gave three kisses to the girl’
c. Sergio les da el beso de despedida a las muchachas.
‘Sergio gave the goodbye kiss to the girls’
In contrast, LVCs referring to a state or change of state do not usually accept an article,
as in the contrast between (3a) and (3b), except when the event is qualified, as in (3c),
and they cannot be quantized (3d) although can accept intensifiers (3e).
3. a
Me dio miedo.
‘I got scared’
b. *Me dio el medio.
Intended: ‘I got scated’
c. Me dio un miedo terrible.
‘I got terribly scared’
d. *Me dio tres miedos.
Literally: (it) gave me three fears.
e. Me dio mucho miedo.
‘I got really scared’
Even though there exist more particular restrictions for some LVCs, this first difference
between types of nouns accounts for many cases in which there seems to be syntactic
restrictions on the noun (cf. Alonso-Ramos 2004). Those restrictions are not arbitrary,
nor are they due to a fossilized syntax, but rather have a semantic orientation.
2.2 Light Verb Constructions are not lexical phrases.
There are other expressions in language that also contain a VP in which the verb does not
seem to contribute to the overall meaning of the sentence, as in:
7
Rodrigo Romero-Méndez
4. My sea-horse kicked the bucket last week.
From a semantic perspective, one cannot make sense of this sentence as a sea-horse
kicking anything, for more than obvious reasons; rather one knows it means that the
animal died. The interpretation of (4) is not related to its parts at all: neither the verb nor
the direct object NP contributes to the meaning of the sentence. For a competent speaker
to interpret (4) is necessary to learn the whole phrase as a lexical item. In contrast, in
LVCs, the meaning of the whole construction is related to the meaning of its parts:
traditionally the meaning has been attributed to the noun, although here I will argue that
the verb also contributes.
Related to this, there is another semantic difference. In order to understand idioms,
one needs to know the cultural context in which they were created, which may result
more or less transparent, as in (5a) if one knows the effect of giving free rein to horses, or
rather obscure as in (5b), which has nothing to do with falling or twenties.
5. a
b
El señor Arbusto le da rienda suelta a su imaginación.
‘Mr Arbusto gives free rein to his imagination.’
Nunca le
cayó
el veinte.
never to.him fall.3SG.PAST the twenty
‘He never got/understood it.’ (Literally: the twenty never fell to him.)
The second semantic difference is that each idiomatic expression has a unique cultural
context that originated it. As will become evident in the following sections, even if the
origin of LVCs relies on metaphors, there is a semantic process that explains their
interpretation of existing LVCs and permits the creation and interpretation of new ones.
Idioms of the kind shown in (5) are lexical phrases, i.e. non-analyzable phrases that
behave as a single lexical unit. This means that one cannot create more idioms with caer
‘to fall’ or veinte ‘twenty’ from the ones in (5). In contrast, one of the main
characteristics of LVCs that I want to emphasize is that they are semi-productive: from
8
Spanish Light Verb Constructions
(6a) one can create other LVCs in (6b) and (6c). Crucially, the examples in (6b) and (6c)
are systematically related to the example (6a).
6. a. Me dio frío.
to.me gives cold
‘I got cold’
b. Me dio calor/hambre/sueño.
to.me gives heat/hunger/sleep.
‘I got hot/hungry/sleepy’
c. Me dio/tengo/me agarró frío.
to.me gives/(I) have/ to.me grabbed cold
‘I got cold / I am cold / I got cold’
But perhaps the main syntactic difference between LVCs and lexical phrases is that the
former can be modified while the latter are syntactically frozen. Consequently, it is not
possible to pluralize (7a) into (7b), change the article (7c) or quantify the noun (7d). As
one can expect, it is not possible to substitute the direct object (7e) or to cleft it (7f).
7. a. Sergio nomás les da el avión a las muchachas.
Sergio just to.them gives the plain to the girls
‘Sergio just patronizes the girls
b. *Sergio nomás les da los aviones a las muchachas.
Sergio just to.them gives the.pl plaines to the girls
c. *Sergio les da un avión a las muchachas.
Sergio just to.them gives a plain to the girls
d. *Sergio nomás les da muchos aviones a las muchachas.
Sergio just to.them gives many plain to the girls
e. *Sergio se lo da a las muchachas.
Sergio just to.them it gives to the girls
f. *El avión es lo que Sergio les da a las muchachas.
the plain is what Sergio to.them gives to the girls
In contrast, in LVCs the P-noun could accept all the previous modifications. In section
2.1 I provided examples that show under which conditions it is possible to modify the
noun. In the following section, I will present the characteristics of the noun as the direct
object.
9
Rodrigo Romero-Méndez
Of course even lexical phrases might have open slots as well, and as a result one can
say that anyone could kick the bucket. To what extent idioms in general have more open
slots and a less restricted syntax is a matter of another discussion.4
As important as the difference between lexical phrases and LVCs may seem, it is not
commonly drawn for Spanish. Even in the most comprehensive Spanish grammar LVCs
are rendered as lexical phrases by scholars (Piera & Varela, 1999; Gómez Tórrego
1999)5, let alone in more traditional grammars such as the one by the Spanish Royal
Academy.
2.3 Number of arguments in LVCs
As mention above, one of the earlier observations made about LVCs was that the number
of arguments seems to depend on the noun, not on the verb. This might be the case for
cases like the following example, where in the (8a) there is one argument, the bus, in
addition to the P-noun, while (8b) there are two arguments, the bus and the Heribeto’s
article.
8. a
El autobús hizo cinco paradas
the bus made
five
(antes de llegar a su destino final)
stops
‘The bus stopped five times (before reaching its final destination)’
b
Francisco le
Francisco
hizo
dat
críticas al
artículo de Heriberto.
made criticisms to;the
article
of
Heriberto
‘Francisco criticized Heriberto’s article’
As pointed out, it became evident that the verb was not devoid of meaning because it
might impose some selectional restrictions (Grimshaw and Mester, 1988). In addition,
scholars soon realized that there are some other cases in which the LVC has arguments
that are not licensed by the noun, and then the idea of semi-light verb constructions came
4
But see Riehemann (2001) and the references therein. She would also argue that in some idioms are
related to their parts.
5
This is stinking if one thinks that this grammar is quite comprehensive: three volumes with more than a
thousand pages each.
10
Spanish Light Verb Constructions
around (Di Sciullo and Rosen 1990, Pelletier 1990). In this case, the argument structure
for the LVC would not be entirely empty but rather it would have an argument specified,
while the other is transferred from the noun. For example, in (9a) the LVC has two
arguments, in addition to the P-noun. However, the P-noun can only license one
argument by itself, not the two arguments of the LVC, as shown in the contrasts between
(9b) and (9c).
9. a
El olor a
the pain of
pan recien horneado me
bread freshly baked
to.me
dio hambre
gave
hunger
‘The smell of freshly baked bread made me get hungry’
b) Mi hambre
‘My hunger’
c) *Mi hambre de/por/hacia/con respecto al olor a pan’
‘My hunger of/for/to/with respect to the smell of bread’
For Italian, Samek-Lodovici (2003) suggested that the number of direct verbal arguments
of an LVC depends entirely on the verb. Thus, LVCs with one argument (in addition to
the P-noun) are formed with fare ‘to do’ because it is monotransitive, while LVCs with
two arguments are formed with dare ‘to give’ in virtue of being ditransitive. Spanish
LVCs contrast with this situation since it is possible to have LVCs with two arguments
using hacer ‘to do’ (10a); conversely, there are LVCs with one argument with dar ‘to
give’ (10b). Therefore, the number of arguments does not depend on the argument
structure of the verb, since (10b) would be impossible, with an otherwise ditransitive
verb.6
10. a) No me
not to.me
hizo gracia su chiste.
made fun
his
joke
‘I didn’t find his joke funny’
Arguments of the LVC (in addition to P-noun): me ‘to me’ and su chiste ‘his
joke’.
6
Of course, it is possible to say that in (10b) dar is not the same verb as in Ricardo me dio su libro
‘Ricardo gave me his book’. However, I will argue that is not necessary to postulate homophonous forms
for the verb.
11
Rodrigo Romero-Méndez
b) Ricardo dio de saltos.
Ricardo gave
of
jumps
‘Ricardo jumped’
Arguments of the LVC (in addition to P-noun): Ricardo.
Even in recent papers (Alba Salas 2002, Alonso Ramos 2004) it has been argued that at
least one argument for the LVC has to come from the noun. The underlying idea is that
there must be argument transference (or a similar mechanism) as a condition for having
an LVC. However, this analysis would not predict cases in which no argument depends
on the noun, as will be exemplified below.
There are other LVC in which the nominal alone does not allow any kind of
argument. As shown in (11), the possession of the nominal calor ‘heat’ results odd (11bc), and it does not accept the cause of the heat either (11c). However, in (11a) there are
two arguments related to the state of getting hot, the cause esa caminata ‘that walk’ and
the experiencer me ‘to me’.
11. a) Esa caminata me dio calor.
‘That walk made me get hot’
b) ?Mi calor
‘My heat’
c) ?El calor por parte de Juan
‘John’s heat’
d) ?El calor de la caminata
‘The heat of the walk’
Finally, let us consider the following case, (12), in which the same verb and noun
combination produces different interpretations as well as a different number of
arguments. In the first case (12a), there is a direct cause, the draft, triggering the change
of state. In the second case (12b), in contrast, there is no cause. One cannot argue that the
cause has been elided because (12b) can be perfectly uttered without any anaphoric or
exophoric referent.
12
Spanish Light Verb Constructions
12. a. Cierra la ventana porque el chiflón me dio frío.
‘Close the window, because the draft made me get cold.’
b. No sé por qué, pero me dio frío.
‘I don’t know why, but I got cold’
This contrast has not been previously discussed in the literature, and certainly presents a
challenge for all of the other theories, since both constructions seem related: (12b) is the
anti-causative version of (12a). It is pertinent to remark that the meaning of (12b) is not
“it makes me cold” or “It makes me fell cold”; in other words, they are not missing the
cause, nor is it pragmatically inferred; it is simply not part of the meaning.
Both in (11) and (12) it would be problematic to argue that the arguments of the
noun have been projected to the whole construction (either via argument transference or
other similar mechanism). As I will argue in this paper, argument transference is a
byproduct of having a syntactically complex predicate, in which part of the semantic
content of both the noun and the verb is combined. Therefore, there can be LVCs in
which the arguments are not attributed to either the noun or the verb, or where a
seemingly identical construction displays different argument structures. In order to argue
for complex predicate formation, it will be necessary a theory that allows for semantic
decomposition. Summarizing, even though in many cases it seems that the noun transfers
its arguments to the verb, it is rather the syntactically complex predicate that licenses the
arguments.
13
Rodrigo Romero-Méndez
2.4 LVCs and their syntactic status
One of the main characteristics of LVCs is that the P-nominal is a verbal argument. In
most cases, it is the direct object (13a), but it can also be a prepositional object (13b)7 or
the subject in a handful of them (13c).
13. a) Los policías le
the
dieron golpes a
policemen to.her gave
beats
la viejita.
to
the old.lady
b) Los policías agarraron a golpes a
la viejita.
‘The policemen beat the old lady’
the
policemen grabbed
to beats
to
the old.lady
‘The policemen beat the old lady up’
c) A la viejita le
to
entró un susto tremendo
the old.lady to.her entered a
scare tremendous
‘The old lady got tremendously scared’
In this section, I will focus on cases like (13a), where the P-noun is part of the direct
object. The question here is whether the P-nominal really behaves as the direct object or
rather it has a more restricted distribution that might suggest that it has a different
syntactic status. The question is relevant for two reasons. First of all, from a broad
theoretical perspective, there is a dispute in some languages about whether in LVCs the
object is incorporated into the verb (cf. Ahn 1990, Dubinsky 1990, Dubinsky 1994,
Grimshaw & Mester 1988). For example, for Japanese, Ahn (1990) states that the verb
suru ‘to do/make’ is a heavy verb when assigns accusative case. Only the cases when
suru ‘to do/make’ does not assign accusative case are true LVCs. Therefore, it is
necessary to explore whether the P-noun in LVCs in Spanish is argument-like or rather it
seems demoted to another syntactic role.
Second, Alonso Ramos (2004) assumes a rather undefined position with respect to
the syntactic status of the P-nominal in Spanish LVCs: they are in the object position but
7
Notice that whether prepositional objects or this kind are arguments or not is a different issue.
14
Spanish Light Verb Constructions
are not exactly like other objects because the whole LVC does not have a phrasal
structure.
Here I argue that the NP containing the P-noun has the same syntactic status as any
other NP in object position. The noun in an LVC can be substituted by the accusative
pronoun in (14a-b), just like with other direct object. As shown in the previous section,
this would not be possible for a lexical phrase. On the other hand, DOs in a leftdislocated position also trigger the use of the DO pronoun, and as one might expect a Pnoun in an LVC behaves just as other DOs, as shown in (14c).
14. a. Le dije que si me seguía molestando le iba a dar una patada
y se la di.
‘I told him that if he kept teasing me I was going to kick him and I kicked him.’
DO pronoun la co-referent with una patada ‘one kick’.
b. A: ¿Le diste un beso a la Julia?
‘Did you kiss Julia?.’
B: No, no se lo di.
DO pronoun lo co-referent with un beso ‘one kiss’.
‘No, I didn’t.’
c. La patada yo se la di.
DO pronoun la co-referent with la patada ‘the kick.’
‘I KICKED him.’
The three previous tests are among the most common diagnostics to determine whether a
NP is a direct object or not, and in LVCs the NP containing the P-noun passes those texts.
The other well known test for DOs is passivization. Contrary to Alonso Ramos (2004),
LVCs can be passivized, as in (15).
15. Todos los golpes fueron dados por la espalda y los costados
‘(Someone) was hit on the back and sides’ (lit. ‘All the hits were given on the back
and sides’) Example from Google:
http://www.infanciayjuventud.com/argenhoy/bsas/baires5.html
Alonso Ramos (2004) points out that P-nouns in LVCs have many syntactic restrictions,
which, as shown, is not the case. Certainly not in all cases the P-noun can be substituted
15
Rodrigo Romero-Méndez
by the DO pronoun, as in the LVC in (16a). However, this is also the case for other bare
nouns, as in (16b) which is clearly not an LVC.
16. a) Pensé que me iba a dar sueño y me (*lo) dió.
‘I thought I was going to get sleepy, and I got (sleepy)’
b) Pánfilo quería comer fruta y (*la) comió.
‘Pánfilo wanted to eat fruit and he ate’
It is important to say that in Spanish, not all types of DOs seem to share the same
properties, and thus some fail to pass the tests for DOs. But this is independent of being
or not an LVC.8
When the P-noun appears in subject position or as part of a prepositional object
(13b-c) above, one could argue more or less along the same lines as I did for direct
objects: they are just as other subject or prepositional object NPs.
2.5 LVCs, auxiliaries and Clause Union
There are two different issues to be discussed in this section. First, for Spanish I will
restrict the use of LVC for V+N constructions, in opposition to V+V complex predicates.
Linked with that, I will argue than in Spanish LVCs are different from auxiliary
constructions and restructuring constructions, restricted to V+V constructions.
In the literature, verbs in LVCs are considered akin to auxiliary verbs or a subclass
of them (Alonso Ramos 2004; Di Sciullo & Rosen 1990; Rosen 1990; Catell 1984). Very
frequently, this goes hand with hand with the fact that in many cases the terms “light verb
construction” and “light verb” have been used for V+N and V+V constructions.
So, different authors have proposed that V+V also count as LVCs (Mohanan 1997
for Hindi; Butt & Geuder 2001 for Urdu and Wagiman, Australian; Pensalfani 2004 for
8
In terms of RRG, only in M-transitive predicates, but not M-intransitive, the direct object satisfies all the
tests for DOs.
16
Spanish Light Verb Constructions
Kalam, Trans-New Guinea, and Jungulu, Australian). A similar phenomenon has been
described for Australian English, in cases like have a read, where the verb to have is
combined with the a V-infinitive construction (Wierzbicka 1982).
17. a
Nyinda-bili-rni
bundurru ukukbili-wunyu-nu.
DEM(m)-dual-ERG food
wrap-3dual-did
‘Those two wrapped the food’ (Pensalfini 2004: 370)
b Raam
has
padaa
Ram-Erg laugh
fall.down-perf-Masc
‘Ram burst out laughing’ (Mohanan 1997:432)
Jingulu (Australian)
Hindi
The distinction between “light verbs” and auxiliary verbs, or the distinction between
verb-noun constructions and verb-verb constructions (which is in principle a different
matter), could be only a terminological issue. However, according Butt & Geuder (2001)
in Urdu auxiliary verbs are different from LVs, even in V+V constructions. In addition,
they acknowledge that for English LVCs are restricted to N+N constructions.
For the particular case of Spanish, there is an obvious difference: V+V and V+N
constructions have different syntactic (and even semantic) properties, and so one should
regard them as auxiliary verb constructions and LVCs, respectively.9 It is not obvious to
me why one should conflate those categories, except for the fact that it has been
traditionally considered that in both the auxiliary verb and the “light verb” are devoid of
meaning (position that is not supported here). One could concede that auxiliary verbs and
“light verb” frequently express aspectual meanings, but that is not enough for equating
them (as Alonso Ramos 2004 suggests), as it would amount to saying that any word with
aspectual meaning should be regarded as auxiliary on similar notional grounds. In other
words, auxiliary verbs should be defined as a subclass of verbs in terms of their
9
Of course, in Spanish it would be also necessary to distinguish the stage and individual level copulas estar
and ser.
17
Rodrigo Romero-Méndez
morphosyntactic properties and not based on the meaning of the word regardless of the
syntactic properties.
For Alba-Salas (2002), it would be possible to conflate auxiliary constructions and
LVCs because he considers that the event denoting noun in an LVC is syntactically a
verb and because in both cases the predicative element displays a “deep subject” (or a 1relation in RG). The problem is that there is no independent motivation for these
suppositions move outside Relational Grammar.
For Alonso Ramos (2004) the analogy between auxiliaries and light verbs comes
with the fact that the noun alone cannot act as syntactic predicate, even though it is the
main predicate at the semantic level. As she says, in constructions like dar una paliza ‘to
beat someone up’, “the support verb is nothing but a lexical tool, used with
morphological and syntactic purposes to allow the construction of the sentence”.10 The
main problem at this point is to make LVCs and auxiliary constructions similar at the
syntactic level so that one can treat auxiliary verbs and “light verbs” as members of the
same category. And it is a problem not just because one class takes NP as complements
and the other combines with verbs, but also because they exhibit different syntactic
properties.
There are two important differences between auxiliary verbs and “light verbs”. It was
shown in the previous section that P-nouns are in many respects just like regular DOs.
However, auxiliary verbs, specially restructuring verbs, do not behave like DO taking
verbs. In other words, volar in (18) cannot be substituted by the DO pronoun; rather, in
addition to lo, it is necessary to use the verb hacer ‘to do’ a proform.
10
The translation is mine, the original quote goes as follows: “El verbo de apoyo dar no es más que una
herramienta léxica, empleada con fines morfológicos y sintácticos para permitir la construcción de la
oración”
18
Spanish Light Verb Constructions
18. a
Yo puedo volar.
‘I can fly’
b. *Yo lo puedo
I it can
c. Yo lo puedo hacer.
‘I can do it’
For Di Sciullo & Rosen (1990) and Rosen (1990) auxiliaries and “Light Verbs” are
similar in that in both cases there is a substitution of the unspecified variables of the first
verb for the argument structure of the second predicate. Let us assume that argument
substitution captures what happens in LVCs in Spanish and let us think only in terms of
what happens at the level of the argument structure. Then, the analysis might seem
correct. In fact, similar analyses have been proposed for Hindi (Mohanan 1997), and
Italian (Samek-Lodovici 2003).
According to this view, when a verb like suele ‘used to VERB’ in (19a) combines
with another predicate, it would have an empty conceptual structure, and therefore it
needs the arguments of the other predicate in order to have any projected arguments.
Then, the argument structure of both verbs merges in a single predicate, like in the case
of LVCs. On of the main evidences for this merge is clitic climbing, as depicted in the
contrast between (19b) and (19c): in (19b) the DO pronoun is an enclitic, but in (19c) it
appears now as a proclitic of the auxiliary verb.
19. a. Ponciano solía leer la carta.
‘Ponciano used to read the letter’
b. Ponciano solía leerla.
‘Ponciano used to read it’
c. Ponciano la solía leer.
‘Ponciano used to read it’
The problem then is that in clear cases of restructuring verbs, the auxiliary and cannot be
separated from the other verb, therefore, it is not possible to have cleft constructions, as
19
Rodrigo Romero-Méndez
illustrated in (20). In contrast, I have shown that in LVCs it is possible to dislocate the
NP containing the P-noun from the verb (14c above).
20. a) Pancho se quiere comer la guajolota.
‘Pancho wants to eat the subway’
a´) *Comer la guajolota es lo que Pancho se quiere.
Intended: ‘To eat the subway is what Pancho wants’
b) Carlitos suele ir al Chopo.
‘Carlitos used to go to Chopo’
b´) *Ir al Chopo es lo que Carlitos suele.
Intended: ‘Carlitos used to go to Chopo’
So, the moral for this section is that LVCs are different from auxiliary verbs mainly
because LVCs take NPs as arguments. The resulting construction, the LVC, behaves
syntactically different from auxiliary constructions. Even if one believes that there is
argument substitution, which I do not, there is no clause union because there are not two
clauses, just one: in the case of LVCs one is not dealing with a complex clause.
Even though I argue that there is co-predication (both the noun and the verb
contribute to the LS), in Spanish LVCs the noun is still an argument of the LV, not a cohead with it (or more precisely, there is no nuclear co-subordination, see Van Valin &
LaPolla 1997, Van Valin 2005).
2.6 Light verbs and Collocations
Alonso Ramos (2004) defines LVCs in terms of collocations. A collocation is defined
technically as “a semi-phrasal expression formed by two lexical items L1 and L2 in which
L2 is taken in a somehow arbitrary mode to express a given sense and/or syntactic role in
function of L1’s choice” (Alonso Ramos 2004:20).11 In terms of argument structure,
11
Original text in Spanish: “Una colocación es una expresión semifraseológica formada por dos unidades
léxicas L1 y L2 n donde L2 es acogida de un modo (parcialmente) arbitrario para expresar un sentido dado
y/o un papel sintáctico en función de la elección de L1”
20
Spanish Light Verb Constructions
Alonso Ramos assumes that the verb “borrows” some of the arguments form the verb, a
solution rather similar to argument substitution or argument transference.
She observes that in an LVC the verb does not seem to be freely chosen, but rather it
depends upon the P-noun. According to her analysis, if one wants to express a predicate
relative to the concept ‘desire’, in Spanish one might use a verb (desear ‘to desire’), an
adjective (deseoso ‘eager’) or a noun (deseo ‘desire’). In case one chooses to use the
noun, then all one needs to know is that it has to be combined with the verb tener ‘to
have’. If one wants to talk about some hitting someone else, and if one decides to use a
noun, then one can combine the noun golpe ‘hit’ with different verb, namely dar ‘to
give’, asestar ‘to deal’ or recibir ‘to receive’. I do not think that using a verb, an
adjective or a noun (in a LVC) are interchangeable instantiations of the same abstract
predicate, and the same goes for two different LVCs with different verbs and the same Pnoun. In addition, I also reject the idea that the verbs involved in LVCs as mere syntactic
support for the main meaning, as she claims. However, there is something true in
thinking that not any noun can combine to any verb to form an LVC, or, in other words, it
there is some degree of idiosyncrasy involved in LVCs. Then, there should be some
information regarding which noun combines with which verb in the lexical information
for both of them, something that she emphasizes. I will address this issue in section 3
(particularly §§ 3.4 and 3.5)
Finally, LVCs cannot be viewed as undergoing a process of grammaticalization,
something that one could expect from semi-phrases. So, for example, the combination of
patada ‘kick’ with dar ‘to give’ as in Mario me dio una patada ‘Mario kicked me’ has
21
Rodrigo Romero-Méndez
been documented since the 15th century.12 Butt and Geuder (2001) have argued
something similar for Urdu.
Summarizing, even thought at some level LVCs seem to be a multiword lexical unit,
as what seems to be the main predicate is extracted from both the verb and the noun (or
just the noun, in Alonso Ramos’ view), treating them as multiword lexical units
contradicts the syntactic properties examined in the previous sections.
2.7 Some regularity: families of LVCs
In the previous section I mentioned that LVCs take a state or event denoting noun as
argument. That is the first classification one can make. In addition, if there is an LVC
with a noun from a semantic field, it is often possible to create more LVCs with nouns
from the same semantic field. Knowing that, it is possible to identify different “families”
of LVCs according to the type of noun, as exemplified in (21).
21. Some families of LVCs with dar.
Internal experiences: Dar calor ‘to get hot’, dar frío ‘to get cold’, dar sed ‘to get
thirsty’, dar hambre ‘to get hungry’...
Diseases: Dar gripe ‘to get the flu’, dar migraña ‘to get migraine’, dar hepatitis ‘to
get hepatitis’…
Emotions: dar miedio ‘to get scared’, dar pavor ‘to dread’ dar coraje ‘to get angry’,
dar envidia ‘to get envy’...
Hitting: Dar patadas ‘to kick’, dar golpes ‘hit’, dar puñetazos ‘to punch’…
Communication: Dar orden ‘to emit an order’, dar opinión ‘to emit an opinion’, dar
un informe ‘to inform’...
An important fact that has not been highlighted in previous studies is that it is possible to
create new LVCs based on other nouns of the same semantic type. For example, one
could make up the noun piernazo, which does not exist but is a perfect derivation from
12
Documented in J. Pérez de Moya, Philosofía Secreta, as displayed in the web searchable corpus Corpus
del español (http://www.corpusdelespanol.org/).
22
Spanish Light Verb Constructions
pierna ‘leg’,13 and would have the meaning of ‘violent contact with/at the leg’, just as
other nouns derived with -azo from a body part mean ‘violent contact with/at noun’:
manazo from mano ‘hand’ means ‘hitting with hand’, rodillazo from rodilla ‘knee’
means ‘hitting with knees’, etc. If someone says (22), it would mean either that someone
hit Roberto on the leg or that someone hit him with their leg. This is a productive part for
LVC formation that includes body parts, but also nouns such as palos ‘stick’, leños
‘firewood’, pedrada ‘hitting with stone’, and many other objects that one can hit with or
throw.
22. Le dieron un piernazo a Roberto.
‘Roberto was hit at/with the leg’
Conversely, a noun can be combined with more than one verb, creating a different type of
“family”. For example the noun patada could be combined with dar ‘to give’, agarrar ‘to
grab’, propinar ‘to give’, recibir ‘to receive’, asestar ‘to deal’, pegar ‘to stick’, tirar ‘to
throw’, largar ‘to give’, disparar ‘to shoot’, chutar ‘to throw’, descargar ‘to
unload/deal’, aplicar ‘to apply’, and plantar ‘to plant’. Roughly all the nouns that can be
combined with dar ‘give’ to form LVCs for hitting can be combined with these other
verbs to express related meanings; sometimes they have a very similar meaning but
belong to another dialect or register (such as propinar literally ‘to give medicine’ instead
of dar ‘to give’: le propinó un coscorrón ‘he knocked him on the head’), sometimes they
express the reverse meaning (such as recibir ‘to receive’: recibí una puñetazo ‘I received
a punch’), or even sometimes they express an intensified meaning (such as agarrar ‘to
grab’: la policía lo agarró a golpes ‘the police beat him up’).
13
I could not find a single example of piernazo in the CREA (Corpus of the Royal Spanish Academy).
23
Rodrigo Romero-Méndez
Perhaps the “hitting family” is one of the more abundant realms for LVCs, but it is
not the exception. Other nouns can be combined with more than one light verb to
highlight different aspects related to the noun. Some of these cases appear below.
23. Family or LVCs with the same noun.
Dar calor ‘to get hot’, tener calor ‘to be hot (=feel hot)’, hacer calor ‘to be hot
(=there to be hot)’...
Dar hambre ‘to get hungry’, tener hambre ‘to be hungry’, hacer hambre ‘there exists
hunger’, by analogy with hacer calor...
Dar golpes ‘to hit’, agarrar a golpes ‘to hit (over and over and with intention)’,
propinar golpes ‘to hit’, recibir un golpe ‘to receive a hit’, asestar un golpe ‘to
hit’...
Darse un baño ‘to have a bath’, tomar un baño ‘to take a bath’...
It is not my intention to provide a taxonomic or a comprehensive list of LVCs. The point
here is that there are more regularities than people usually assume. One important
characteristic of LVCs is that often times the same P-noun can select different verbs. This
fact has not always been highlighted and languages may differ as to how many verbs a
given noun can combine with. Alonso Ramos (2004) had noticed this, but at the same
time, she claims that the lexical content of the verb is already included in the lexical
content of the noun. Following her line of thought, in (24) the idea of kicking is already
contained in the noun patada ‘kick’ and so the verb to give does not provide any
meaning.
24. Mario le dio una patada a Miguel.
‘Mario kicked Miguel once’
Then, the fact that the same noun can be combined with different nouns becomes a mere
lexical curiosity. In other words, if one claims that the verb does not contribute to the
LVC, there is no reason as to why the same noun can be combined with many verbs to
form semi-phrasal (i.e. semi-lexical) constructions, in Alonso Ramos’ account. Related to
this, in her view the selection of the verb is arbitrary, not related to its meaning.
24
Spanish Light Verb Constructions
2.8 Summary
In this section I have argued that there are two main groups of LVCs depending on the
type of P-noun, and that these groups have different syntactic characteristics. I also
indicated that in some cases the arguments are not licensed by the P-noun, and in fact it is
possible to think that in general the arguments structure of the LVC is the result of the
“merge” of the meaning of both the noun and the verb. In other words, it is the
construction itself, not its components, that licenses the arguments. Finally, I have also
provided evidence for claiming that in syntactic terms the verb and the P-noun are not coheads (there is no nuclear co-subordination). Rather, the P-noun’s NPs share the syntactic
properties of other NPs in object position. In other words, the LVC has phrasal
properties, although in some cases there are some idiosyncratic restrictions, which is why
one could think of them as collocations. Finally, I argued that LVCs can be grouped into
“families”: there is regularity within those families but there is a great deal of
idiosyncrasy across families.
3 Light verb constructions in Role and Reference Grammar
3.1 Semantic Decomposition in RRG
In this section, I will explain some basic principles for semantic decomposition in Role
and Reference Grammar. In RRG, the semantic representation is based on a Vendler’s
(1967) classification of verbs. The original verb classification in RRG (Foley & Van
Valin 1984; Van Valin 1993; Van Valin & LaPolla 1997) distinguished states,
25
Rodrigo Romero-Méndez
achievements, accomplishments and activities. This classification is well known and I
will not elaborate here, but we can see the list in (25).14
25. a.
b.
c.
d.
States: John is sick.
Achievements: The balloon popped.
Accomplishments: The snow melted.
Activities: John runs.
Over the past years, two other categories have been added: semelfactives (Smith 1997)
and active accomplishment. In the first case, it has been observed that even though
achievements and semelfactives are very similar in that both are punctual, the former
lacks a resulting state: The light flashed (semelfactive) does not entail The light is
flashed; but The window shattered does entail The window is shattered. On the other
hand, the incentive for having active accomplishments in the repertoire is the observation
that activities behave in a different way when they have an end point, like in (26b). So, if
one says Pete is walking, then one can truthfully say Pete has walked; but if one says
Pete is walking to the store, one cannot truthfully say Pete has walked to the store (in that
particular occasion) since he has not reached the store yet.
26. a. Semelfactive: Richard glimpsed Roselyn’s smile.
b. Active accomplishment: Pete walked to the store.
All the previous six verbal classes have, in addition, a causative counterpart. So for
example, the beer is cold expresses a state predicate but the fridge cooled the beer
expresses a causative state: the fridge causes the beer to be cold.
There are two classes of “words” in the metalanguage used for semantic
representations in RRG: predicates and operators. The formal representation using
predicates and operators is called Logical Structure. Both states and activities are taken as
14
I will present the basic notions here, but for a deeper and more detailed description see Van Valin &
LaPolla 1997 and Van Valin 2005. Furthermore, the labeling changed from Foley & Van Valin (1984) and
Van Valin (1993) to Van Valin & LaPolla (1997), and then it was adjusted in Van Valin (2005).
26
Spanish Light Verb Constructions
basic in RRG, and they are represented using just predicates. States receive the simplest
representation, they are not marked by any special operator and are represented just with
a bare predicate, as shown in (27a); in activities, it is necessary to add the predicate do´,
in addition to the predicate indicating the “kind” of activity (i.e. walking vs. running), as
illustrated in (27b).
27. a.
a'.
b.
b'.
States: predicate´ (x), where x is an argument.
The beer is cold: cold´ (beer).
Activities: do´(x, [predicate´(x, y)]), where x and y are arguments.
My buddy is drinking beer: do´ (buddy, [drink´ (buddy, beer)])
In order to obtain the correct decomposition for activities, one has to do two things: first,
make sure to have the predicate with its arguments, [drink´ (buddy, beer)] in (27b').
Since drink does not refer to an state, drink´ (x, y) cannot be a complete LS per se;
rather, it is just a part of the complete representation that always occurs as one of the
arguments of do´(x, [pred´(x)]), where pred´(x) represents the other part of the lexical
meaning that lets us distinguishing walk from run, or from any other activity, for that
matter.
All other types of predicates are represented using the second class of “words” used
in the metalanguage: operators. In this respect, achievements have the operator INGR,
accomplishments BECOME, and semelfactives SEML, all of them combined with the
respective predicate. Consequently, in order to represent the snow melted, one has to add
BECOME to the predicate to form: BECOME melted´ (snow).15
In contrast, causatives have a slightly more complex form, depicted in (28a). So,
turning back to our previous example, in (28b), the fridge does something that causes the
15
Notice that do´, the marker for activities in do´(x, [pred´(x)]), does not have the same status as other
predicates, and rather behaves as a semantic operator. We do not mark it in capital letters, as any other
operator, since DO is reserved for marking agents, in opposition to effectors (see VV&LP 1997:118 ff.).
On the other hand, do´(x, ∅) represents an unspecified activity, and so in the LS it does resemble a
predicate.
27
Rodrigo Romero-Méndez
beer to become cold. Since one does not know or one does not specify what the fridge
does, one just represents an unspecified activity: do´(fridge, ∅).
28. a. Causatives: α CAUSE β, where α, β are LSs of any type
b. The fridge cooled the beer: [do´(fridge, ∅)] CAUSE [BECOME cold´ (beer)]]
Finally, there are lexical rules signaling the relation between activities (Pit drank beer)
and active accomplishments (Pit drank the beer). This is particularly relevant for verbs of
consumption and creation. The lexical rule is stated as follows (From Van Valin 2005):
29. Creation/consumption verbs:
do´ (x, [pred1 ´ (x, y)]) <—> do´ (x, [pred1´ (x, y)]) & INGR pred2´ (y)
where pred2´ is of the type of creation (exist´) or consumption (consumed´).
Even though RRG does not have a defined set of semantic primitives, there have been
several attempts for a more refined semantic decomposition (Van Valin & Wilkins 1993,
Mairal & Faber 2002, Mairal & Faber 2005). For example, Van Valin & Wilkins (1993)
provide the decomposition in (0) for remember. When needed, I will use a semantic
decomposition along those lines.
30. BECOME think.again (x) about something.be.in.mind.from.before (y)
In the next sections, I will explain how the formation of LVC at the LS occurs from two
different lexical items. In particular, how the event denoting noun is combined with the
so-called light verb in order to create the LVC.
3.2 Initial analysis of Light Verbs
One of the motivations for having do´ for actions comes from Basque. As Van Valin and
LaPolla (1997) and Levin (1989) point out, in this language verbal expressions that are
usually lexicalized as intransitive activity verbs in languages like English are created by
combining a noun with the verb egin ‘do, make’ as one can see in (31). Furthermore,
28
Spanish Light Verb Constructions
Levin (1989) indicates that in these cases the noun is non-referential, but it picks out the
action denoted by it.
31. ni-k
lan-∅
egin d-u-t
work-ABS do
3sgABS-AUX-1sgERG
1sg-ERG
‘I worked’ (from Levin 1989:54)
The combination of a non-referential noun meaning ‘work’ plus a verb meaning ‘do,
make’ should be familiar by now. Even though not pointed out by Van Valin and
LaPolla, this is exactly the type of combination that we have seen in previous examples
(as in El autobús hizo cinco paradas ‘The but stopped five times’, in (8a) above). In other
words, RRG already has an initial mechanism for dealing with LVCs: the representation
for the noun fills the second position in the representation do´(x, y).
This solution works for all LVCs that use hacer ‘do, make’. Let us take the LVC in
(32a). One could assume that the meaning of hacer in this case is do´(x, y), where y can
be a predicate; the meaning of llamada ‘call’ would be call´(x). This gives rise to the
representation in (32a’).
32. a. Jacinto hizo una llamada.
‘Jacinto made a call.’
a'. do´(Jacinto, [call´(Jacinto)])
Then, LVCs are the result of combining the meaning of the noun with the meaning of the
verb. This is different from saying that the verb acts only as a syntactic support for the
noun. Notice that the LS for hacer ‘to do’ alone only indicates that there was an activity,
without indicating the type of activity. In the LS for hacer una llamada ‘to make a call’,
the LS for the noun specifies what the activity is. So, LVCs are a special case of cocomposition (Van Valin 2004, Pustejovsky 1995, 1998), where the noun provides the
call´ for the LS. In other words, what is important in the formation of LVCs is that the
noun provides the pred´ in the LS, replacing the one from the verb.
29
Rodrigo Romero-Méndez
This solution carries several advantages. First, it has been pointed out in previous
accounts that the main purpose of LVCs is to combine the argument structure of the noun
with the unspecified variables of the verb. In the presented analysis, the resulting
argument structure is the effect of creating the LS based on the LSs from two lexical
items. Then, in LVCs, as in RRG in general, the argument structure follows from the LS
(see Van Valin and LaPolla 1997; Van Valin 2004, 2005). Second, there is no need for
postulating any control mechanism in LVF: it is a natural result of the resulting LS (cf.
Alba-Salas).
In the previous examples the LS for the noun seems highly similar to the LS of a
verb, but this is in part an illusion due to the fact that there is no need for a further
decomposition in the meaning of the noun. There are other cases in which it is necessary
to decompose the meaning of the P-noun when providing its semantic representation.
One may ask, for example, what tener hambre ‘to be hungry’ really means. As the
translation suggests, it refers to a physical sensation (or proprioception) and not to a
possessive relation. Therefore it would not be appropriate to represent Toño tiene un
acordeón ‘Toño has an accordion’ and Toño tiene hambre ‘Toño is hungry’ with the
same LS because in one case Toño possesses something while in the second case he feels
an internal sensation. Therefore, it is necessary to provide a LS as in (33a’). On a similar
note, the meaning of dar hambre ‘to get hungry’ (lit: ‘to give hunger’) in (33b) has noting
to do with transference of possessor, but rather with becoming hungry, represented as in
(33b’).
33. a. Toño tiene hambre.
‘Toño is hungry.’
a’. feel’ (1sg, [hungry’])
b. Ya me dio hambre
30
Spanish Light Verb Constructions
‘I got hungry already.’
b’. BECOME feel’ (1sg, [hungry’])16
At the intuitive level, it seems appropriate to say that a part of the LS for dar hambre ‘to
get hungry’ should have the predicate feel’, but where does it come from? In order to
answer this, it is necessary to provide a richer semantic representation for nouns.
3.3 The semantics of predicative nouns
It is well known that not all nouns refer to physical entities, like dog, chair or people;
other nouns refer to activities, like destruction (Chomsky 1970; Langacker 1987, Nunes
1993, Pustejovsky 1995, Picallo 1999, inter alia). Furthermore, it is also known that this
last type of nouns has two interpretations: one refers to concrete entities and the other one
to activities. For example, in (34a) the noun refers to the event of reporting, an
interpretation that is selected by the verb, since only eventualities can be extended in
time, as signaled by the verb durar ‘to last’; on the other hand (34b) refers rather to the
physical object containing the information, since only physical objects can be located on
a table.
34. a. El informe duró cinco horas
‘The report lasted five hours.’
b. Casimiro dejó el informe sobre la mesa
‘Casimiro left the report on the table.’
In order to make the distinction between nouns referring to activities and nouns referring
to the result of an activity, the terms process nominal and result nominal are used. This
terminology can be somehow misleading, since it is also applicable to nouns referring to
states. In the following examples (35a) refers to the disease itself whereas (35b) refers to
16
As discussed below, the meaning of dar ‘to give’ as a heavy verb can be decomposed as the causative
counterpart of tener ‘to have’, also as a heavy verb. This does not apply, however, for LVCs. As tempting
as it may seem to say that dar frío ‘to get cold’ is the causative of tener frío ‘to be cold’, it would be
incorrect to use the LVC tener frío in the LS of dar frío. Certainly at an informal level one could use a LVC
in order to understand another one, but this cannot be extrapolated to their LSs.
31
Rodrigo Romero-Méndez
the state of being sick. As Nunes (1993) points out, what is important to note is that
nouns referring to states also have a more concrete interpretation.17
35. a. Descubrieron una nueva cepa de gripe
‘A new flu strain was discovered.’
b. La gripe le duró dos semanas
‘He had the flu for two weeks.’ (lit.: the flu lasted to him two weeks)
This discussion is relevant because LVCs take the process or state sense of the noun, and
not the concrete one. In addition, it is clear that these two interpretations should have a
different representation in virtue of one being a concrete entity and the other one being a
state or a process. This begs the question, how are these two interpretations related? For
Alonso Ramos (2004), this is just reduced to listing the different interpretations, which is
far from being satisfactory, while Alba-Salas (2001) argues that process or state denoting
nouns are in fact verbs, but there is no evidence that supports his claim.
Pustejovsky has suggested that the meaning of nouns includes information regarding
the type of entity (i.e., whether it is a physical object, information, etc.), its components,
and the eventualities that are associated with those entities. This information, organized
in a systematic way, is called the qualia structure (Pustejovsky 1991, 1995, 1998, 2000,
2001).
The qualia structure of nominals has been of incorporated in RRG in recent years
(Van Valin 2004, 2004, Van Valin & LaPolla 1997), although here I will use it without
the formalisms the Pustejovsky uses. In order to explain the qualia structure, take for
example the noun novel. Novels contain narrative (36a) and they come very frequently
contained in books (36b). On the other hand, there are two activities associated with
17
When dealing with nominalizations, we could apply further distinctions, such as agent nominalization,
where the noun incorporates the agent of the verb, like in buyer, Spanish comprador; or patient
nominalizations, where the noun incorporates the patient of the verb, like invention, Spanish invención (see
Nunes 1993, Picallo 1999, Portero 2003), but this further classifications are not relevant here.
32
Spanish Light Verb Constructions
novels, people write them (36d), and people read them (36c). Depending on which of
these activities is picked, one can interpret Julio just finished the novel as meaning that
Julio finished reading the novel based on (36c), or that Julio finished writing the novel
based on (36d).18 So, the noun novela ‘novel’ should have a qualia representation as
follows:
36. Qualia: novela ‘novel’
a) narrative´(y)
b) book´(y)
c) do´(x, [read´(x, y)])
d) do´(x, [write´(y, y)]) & INGR exist´(y) (based on Van Valin 2004)
Following Van Valin and Wilkins (1993), one could decompose activity nouns used in
LVCs as in (37). One could say that an explanation contains information (37a) expressed
in detail and in a clear way (37c) that usually has a physical manifestation (37b). Then, if
someone says La explicación duró media hora ‘The explanation lasted half an hour’, the
even represented in (37c) is picked out, but if she says Te dejé la explicación sobre el
escritorio ‘I left the explanation on the desk’, one interprets that it refers to the physical
object (37b). Therefore, there is no need for saying that the event interpretation of the
noun is a different lexical entry (as Alba-Salas 2002), rather a different par of the qualia
is picked out in the concrete and in the event interpretation, but both are systematically
related insofar they are part of the complex meaning of the noun explicación
‘explanation’.
37. Qualia: explicación ‘explanation’.
a) information´(y)
b) physical object´(y)
d) do´(x, [express.(α).to.(β).in.detail´(x, y)]) CAUSE [BECOME aware.of´(y, z)],
where y=β and z=α
18
It should be noted that Pustejovsky (1995, 1998) assumes that the qualia structure is part of the lexical
meaning. I will assume a weaker position here, more along the lines of conceptual frames (Fillmore 1968).
In other words, the qualia structure contains characteristics associated to the lexical meaning; we could
assume that they are part of it or that they are implied.
33
Rodrigo Romero-Méndez
Similarly, for nouns referring to states, also found in LVCs, the noun can refer to the state
itself as in (38a), but the qualia structure has to also include that a necessary part of the
definition of frío ‘cold’ is that it is something that people feel, as in (38b).
38. a. El frío dañó a las uvas.
‘The cold damaged the grapes’
b. El frío me hizo temblar.
‘The cold made me shiver’
c. Qualia: frío
cold’ (y)
feel’ (x, [cold’])
Thus, it is possible to use the information from the qualia structure of nouns to fill in
parts of the LS of LVCs.
3.4 Verbal schematism and qualia in the formation of LVCs
In section 3.2 I provided the basics for creating LVCs using RRG, but it was necessary to
provide a richer semantic representation for nouns in order to explain where the LS for
cases like dar frío ‘to get cold’ comes from. In the previous section, I explained how that
richer representation might look like and how the meaning of some nouns is
systematically related to their concrete meaning. Crucially, both senses are part of the
complex meaning of those nouns. Now, it is time to revise and expand the proposal for
LVCs for more types of LCVs. In addition, I will show how this analysis easily accounts
for the LVC families presented above.
First, it is necessary to have a look at the LS for dar ‘to give’ as a heavy verb to later
compare it with the LS of LVCs. The LS of the example in (39a) is represented in (39b).
The LS must be interpreted as Isra doing an unspecified activity that caused that I had the
34
Spanish Light Verb Constructions
bottle.19 The argument of do´(x, ∅)... is the initiator of the action whereas the first
argument of ...BECOME have´(y, z) is the receiver and the second argument the
transferred object.
39. a
b
Isra me dio la botella.
‘Isra gave me the bottle ’
[do´(Isra, ∅)] CAUSE [BECOME have´(1sg, botella)]
In the case of the LVC dar frío ‘to get cold’ in (40a), whose corresponding LS in (40b),
one can paraphrase the LS as the ice cream causing something such that I get cold. The
ice cream is represented in a generic activity do´(helado, ∅)..., which does not mean that
the ice cream is doing, willingly, something, but rather that it is involved in an
unspecified activity, presumably being eaten. The predicate feel’ (1sg, [cold’]) comes
from the qualia structure of the noun frío ‘cold’ (38c).
40. a
b
El helado me dio frío.
‘The ice cream got me cold’
LVC: dar ‘to give’ + frío ‘cold’
[do´(helado, ∅)] CAUSE [BECOME feel’ (1sg, [cold’])]20
It is necessary to say that the meaning of those constructions as causative is independent
of the analysis in RRG. In other words, if one was to decide to use a different
metalanguage one still would have to account for that part of the meaning. Notice that
dar ‘to give’ also has a causative meaning in other constructions.
41. a. Maricela le dio de comer a su hijo.
‘Maricela fed her child’ where feed=make.eat
b. [do´(Maricela, ∅)] CAUSE [do’(hijo, [eat´(hijo)])]
19
It is an open question to what extent causatives are implicative, i.e. to what extent to give implies to have.
Park (1995) indicates that in Korean causatives are not implicative (in Van Valin 2005). For Spanish, we
can assume that causatives are implicative.
20
Some of these constructions can be achievements or accomplishments. For example, me di un susto ‘I got
scared’ is most probably an achievement, since the change of state happens without transition. However,
since in RRG accomplishments can be decomposed into PROC & INGR, I will represent all these changes
of state as BECOME (Van Valin 2005).
35
Rodrigo Romero-Méndez
All this can be summarized in (42): in the LVC dar frío, the light verb provides the
lexical template and the predicative noun provides the predicate that is missing in the
lexical template. The noun’s predicate comes from its qualia.
42. LVC: Dar frío
Dar ‘give’: lexical template [do´(x, ∅)] CAUSE [BECOME pred’]
Frío ‘cold’: Qualia Structure: feel’ (x, [cold’]), which replaces pred’ in the LS for
dar.
Then, the LS for LVCs is the result of the co-composition of the LS for the verb and the
LS for the noun. This is done syntactically, not lexically (morphologically). One could
think of this as a special type of coercion (Pustejovsky 1995, 1998, inter alia), as in Mary
finished the book, where its interpretation depends on filling in the information relative to
the activities that one usually performs with books, i.e. writing or reading them. The main
contrasts with LVCs is that in the final LS, the qualia from the noun does not only
provides the missing predicate (i.e. that one feels the cold in (42), but it also replaces the
pred’ from the LS for the verb.21
All the previous discussion can be stated as the generalization in (43).
43. Complex predicate formation in LVCs:
i) the verb should provide the lexical template or part of it, and
ii) the qualia structure of the nominal replaces part of the lexical template provided
by the verb; in particular, the nominal should provide the predicate in the logical
structure (pred’(x) or pred’(x, y)).
One of the main contributions of the light verb is to provide the operators in the lexical
representation, i.e. do´, CAUSE, BECOME, INGR, etc. (cf. Van Valin & LaPolla 1997,
Van Valin & Wilkins 1993, Mairal & Faber 2005).
21
A full comparison between type coercion and LVCs would be necessary, but it goes beyond the limits of
this paper.
36
Spanish Light Verb Constructions
In previous sections I reported that there was a problematic case in which the noun
does not accept arguments but the LVC might have two arguments, as in (11) with the
noun calor ‘heat’ or in (40) with the noun frío ‘cold’, repeated here as (44).
44. El helado me dio frío.
‘The ice cream got me cold’
As mentioned above, this case would be problematic for an account based on transference
of arguments from the noun, since the noun alone licenses no arguments. However, it is
not problematic for a proposal where the arguments do not depend directly on either the
verb or the noun but rather on the LS resulting from both elements. In any case, the lack
of explanation for how the complex predication is formed is the main failure of the other
two main descriptions of light verbs in Spanish (Alba-Salas 2002, Alonso Ramos 2004).
3.5 Semicompositionality and semiproductivity in LCVs
From the discussion in the preceding section one could get the impression that LVC
formation in Spanish is a regular process, in fact a compositional one. On the other hand,
it has been reported that LVCs in Spanish are idiosyncratic constructions, in which the
meaning does not necessarily correspond to its parts. As Alonso Ramos (2004) has
emphasized, in some cases the selection of the verb seems arbitrary. How, then, can one
reconcile these two seemingly contradictory positions? In section 2.7 I argued that one
observes regularities within the “LVC families” described in that section, but a great deal
of idiosyncrasy across families. In this section I will argue that LVCs have a degree of
compositionality and therefore they are semi-productive, but they are as well
idiosyncratic constructions.
It was explained that it is possible to identify some groups of LVC because one the
same verb can be combined with several nouns that share part of the same meaning, and
37
Rodrigo Romero-Méndez
the same noun can be combined with more than one verb variants of the meaning denoted
by the noun. I will explain how this happens with a few semantic fields.
It was mentioned before that some LVCs denoting change of state have a
causative/anti-causative alternation, as in (12), repeated here as (45).
45. a. Cierra la ventana porque el chiflón me dio frío.
‘Close the window, because the draft made me get cold.’
b. No sé por qué, pero me dio frío.
‘I don’t know why, but I got cold’
Not all LVCs denoting change of state allow this alternation, though. Specifically, there
is a whole set of LVCs referring to diseases that do not allow the causative component to
be part of the LS of the LVC: they encode just a change of state. The example in (46a)
indicates that Pancracio got sick, and it does not imply that there is an unsaid cause. It is
not possible to have a noun in the subject position as the cause of the change of state
(46b); the only way to express it is using the periphrastic causative (46c).
46. a. A Pancracio le dio gripe
‘Pancracio got the flu’
b. *La lluvia de la semana pasada me dio gripe.
Intended: ‘Last week’s rain made me get a cold’
c. La lluvia de la semana pasada hizo que me diera gripe.
‘Last week’s rain made me get a cold’
Since all the LVCs with dar referring to diseases behave in a similar way, one should
have a particular representation for them (47). So, this schematic representation
represents the semantics of LVCs related to diseases, and then the formation of LVCs is
regarded with a particular type of constructionist approach.
47. LVC dar + noun qualia: {disease’(x)}
Schematic LS: BECOME be.affected.by.disease(α)´(y)
So far, I have only addressed the LVCs with dar that have a change of state as part of
their meaning. In the case of LCV that denote activities, this group differs radically in
38
Spanish Light Verb Constructions
that LVCs do not take the whole [do´(x, ∅)] CAUSE [BECOME...] part of the LS of dar,
but only the part relative to the activity do´(...), as in (48a-b).
48. a. Jeon no se dio una ducha durante todo el semestre.
‘Jeon did not take a shower during the whole semester’
b. do´(x, [shower´(x, y)])
c. Jeon se hizo una ducha.
‘Jeon made (i.e. built) a shower for himself’
The idiosyncratic part here corresponds to the noun ducha ‘shower’, which denotes an
activity, forming an LVC with dar ‘to give’ and not with hacer ‘to do, make’. A
predicate formed by hacer + ducha can only take the non-process meaning of the noun
and therefore it can only mean to build a shower (48c). But this is true for all dar-LVCs
that have an event denoting noun, which is a big class.
Of course, within the activity denoting LVCs, there are many subgroups, and then it
is necessary to provide a particular schematic representation for each of them. For the
case of LVCs referring to hitting, which are a very productive subgroup, one can
represent (49a) as in (49a’). In this case, is necessary to provide a semantic representation
where part of the meaning is decompose in order to capture the generality. Then, the
schematic LS for this group would be as indicated in (49b).
49. a. Mario le dio una patada al jugador del otro equipo.
‘Mario kicked once the other team’s player’
a’. 〈(1)time 〈do´(Mario, [contact.with.force.between.(foot).and.(β)´(Mario,
jugador)])〉〉 where β=jugador
b. LVC dar + noun qualia: {contact.with.force.between.(α).and.(β)´(x, y)}
Schematic LS: do´(x, [contact.with.force.between.(α).and.(β)´(x, y)])
As explained, it is possible to create new LVCs based on the particularities of the LVC
family. For example, if we create a new noun that we intend to refer to a new disease, it
could be also possible to include it in an LVC. Let us make up a nonsense disease called
39
Rodrigo Romero-Méndez
frontitis (the pseudo Latin would work both in Spanish and English), and let us provide
the definition below, taken from my personal “makeupedia”. Then we could readily use
this new disease in an LVC indicating change of state, just as in any other LVC of this
subtype.
50. Frontitis: Inflammation of the frontal lobe from excess in thinking. If the
inflammation persists it can lead to the complete hypertrophy of the frontal lobe,
causing the patient to become an incurable nerd.
51. a. A Francisco ya le dio frontitis.
‘Francisco got the frontitis already’
b. BECOME be.affected.by.disease(frontitis)´(Francisco)
Summarizing, LVCs are indeed idiosyncratic constructions with respect to two things: i)
the verb that a noun takes (i.e. ducha combines with dar to express the activity of
showering but it is incompatible with hacer to express the same meaning); and, ii) with
respect to what part of the light verb’s lexical template remains for the LVC across
different types of nouns. On the other hand, LVCs are semi-compositional and semiproductive to the extent that once one understands the idiosyncrasy, it is possible to
almost compositionally form the meaning of predicates, and furthermore to create new
predicates.22
3.6 Light verb interpretation vs. heavy interpretation
As the final subsection, let me briefly address the relation between light and heavy use of
verbs. There are some cases in which the construction is potentially ambiguous between
an interpretation as an LVC and another as non-LVC. In the second case, the verb would
be interpreted as heavy verb and the noun would have a concrete interpretation. For
22
Even though it is not possible to treat this issue here, idiosyncrasy is one of the main differences between
LVC formation and coercion as presented in Pustejovsky (1995, 1998). To me, coercion is an operation that
takes place regularly and as an ongoing process at the moment of speech, whereas LVC formation is
idiosyncratic and constructional, in the sense that the construction for specific families and the particular
way in which they resolve the semi-composition is stored in the lexicon.
40
Spanish Light Verb Constructions
example, in the sequence dar un baño, the noun baño ‘bath’ could be taken as meaning
the action of taking a bath or something like ‘where the bath takes place’, i.e. the room or
a container.23 Under the first interpretation, we get the LVC meaning ‘to take a bath or a
shower’ (52b). Under the second interpretation in (52c), the LS for the noun does not
replace part of the LS for the verb, and therefore it has to be interpreted in its literal
sense, as a transference.
52. a. Dale un baño a tu hijo.
b. ‘Bathe your son’
do´(2gs, [shower´(2sg, hijo)])
c. ‘Give a bucket to your son’
[do´(2sg, ∅)] CAUSE [BECOME have´(hijo, baño)] 24
Against having two lexical entries for baño ‘bath’, à la Alonso Ramos (2004), it can be
argued that there is just one lexical item that happens to be polysemous. As we already
know, the two parts of the meaning of the noun come from different parts of its qualia
structure. There is no need, either, for postulating any syntactic derivation for the activity
interpretation of the noun, as Alba-Salas (2002) assumes.
For some V+N constructions, the distinction between LVC and heavy construction is
even more subtle. For example, the noun informe ‘report’ would have in its qualia that it
is information, which can be communicated, and that it can be created (presumably
writing it). Then, the sentence in (53a) can mean that I received the physical object
containing the information or that I received the information via communication, and
therefore that I was informed.
53. a. Me dieron el informe sobre la situación en Atenco.
b. ‘They informed me about the situation in Atenco’
23
At least in some dialects baño means the bathroom or a kind of bucket where parents can bath a baby.
For clarity’s sake, the LS for the interpretation referring to the physical object provided in this section are
a simplified version of more accurate LS.
24
41
Rodrigo Romero-Méndez
[do´(3pl, [express.as.report´(3pl, situación)]] CAUSE [BECOME
aware.of´(1sg, situación)]
c. ‘They gave me the report about the situation Atenco’
[do´(3pl, ∅)] CAUSE [BECOME have´(1sg, informe.de.situación)]
In some other cases, with the verb hacer ‘to do, make’ the ambiguity can be
straightforward, since under the heavy verb interpretation the hacer + noun construction
has the meaning of creating something, but under the LVC interpretation it means to do
something relative to the noun’s qualia. In order to make more explicit the relation
between heavy and light uses of verbs, take the following example:
54. a. Juan hizo la silla
‘John made the chair’
do´(Juan, ∅) & BECOME exist´(silla)
b. Juan hizo el estudio
i) do´(Juan, ∅) & BECOME exist´(estudio)
ii) do´(Juan, [study´(Juan)])
In (54a), the interpretation is just that Juan did something, we do not know what,
resulting in the chair being created. We can make a number of assumptions, depending on
how well we know Juan. We could suppose, for example, that the chair is made of
mahogany and that Juan cut down the tree by himself; or we could suppose that he
bought all the pieces in the supermarket and only assembled them. The crucial point is
that there is an unspecified activity, or set of activities, indicated by do´(Juan, ∅), and
that the chair came into existence, indicated by BECOME exist´(silla).
In (54bi) we could say the same; there are an unknown number of activities that lead
to the creation of the study, the physical object. However, if the interpretation of study is
not as a physical object, but rather as an activity, one would have a light verb
construction. In this case, the activity expressed by the noun is already telling us the
42
Spanish Light Verb Constructions
activity involved. In other words, the pred´ contained in the LS of the process
interpretation of the noun replaces the previously unspecified activity.
This systematic process explains LVC formation but also tells us that the noun and
the verb are the same in LVCs and in non-LVCs.
4 Concluding remarks
In this paper I argued that LVCs are syntactically formed complex predicates, in
opposition to lexically (or morphologically) formed complex predicates, where the
meanings from the noun and the verb are merged into a single LS. As argued in section
2.2, this does not mean that LVCs are lexical phrases, i.e. a combination of words that at
the surface level resembles a phrase (in many cases a VP) but that in fact is a single
lexical item. LVCs, on the contrary have phrasal properties, although depending on the
type of predicative noun, whether it is event denoting or state denoting, the syntactic
flexibility might change (but the restrictions are semantically driven). In addition, it was
shown that when the predicative nominal is in object position, it has more or less the
same syntactic properties that one might expect of any other direct object NP.
It was also pointed out that in previous studies verbs in LVCs have been equated
with auxiliary verbs. Against this view, one can say that auxiliary verbs have rather
different syntactic properties than verbs in LVC, and thus there are no good reasons for
conflating these two categories.
If one takes into consideration that LVCs are complex predicates in the sense that
they are made up from two independent words, one might also want to think of them as
nuclear cosubordination. This view would be mistaken, however, because the predicative
noun is not at the same syntactic level as the verb, but rather it is a verbal argument.
43
Rodrigo Romero-Méndez
Under the view proposed in this paper, verbs in LVCs are just like any other verb,
and, likewise, nouns are just like any other (action) noun. What is special in LVC is the
construction in which they appear, and depending on both the verb and the noun, there
are some idiosyncrasies. In this respect, it was argued that there are two main groups of
LVCs, depending on whether the noun denotes a state or an event. More specifically, it is
possible group LVC in so-called “LVC families”. Then, while there is a great deal of
idiosyncrasy across families, there is also much regularity within them.
As a final point, it is worth emphasizing that the argument structure falls out from
the LS, and so there is no need for a special mechanism such as argument transference or
argument substitution.
5 References
ADGER, David. 2003. Core syntax. A minimalist approach. Oxford: Oxford University
Press.
AHN, Hee-Don. 1990. On Light Verb Constructions in Korean an Japanese. Hajime Hoji
(ed.) Japanese/Korean Linguistics, Vol. 1, 221-237. Stanford: The Stanford
Linguistic Association.
ALBA-SALAS, Josep. 2002. Light verb constructions in Romance: a syntactic analysis.
Ph.D. Dissertation: Cornell University.
ALONSO RAMOS, Margarita. 1998. Étude sémantico-syntaxique des constructions à verbe
support. Ph.D. dissertation, Université de Montréal.
ALONSO RAMOS, Margarita. 2004. Las construcciones con verbo de apoyo. Madrid:
Visor.
44
Spanish Light Verb Constructions
BUTT, Miriam, and Wilhelm GEUDER. 2001. On the (semi)lexical status of light verbs. In
N. Corver and H. can Riemsdijk (eds.), Semi-lexical categories: the function of
content words and the content of function words, pp. 324-370. Berlin/NY:
Mouton de Guyter.
BOSQUE, Ignacio. 2001. On the weight of light predicates. In Features and interfaces in
Romance: Essays in honor of Heles Contreras. Amsterdam: John Benjamins.
BRUGMAN, Claudia. 2001. Light Verbs and Polysemy. Language Sciences 23, 551-578.
CATTELL, Ray. 1984. Composite predicates in English. Syntax and Semantics volume 17.
Sydney: Academic Press Australia.
CHOMSKY, Noam. 1995. Minimalist Program. Cambridge, Mass: MIT Press.
DI SCIULLO, Anna Maria, and Sara Thomas ROSEN. 1990. Light and semi-light verb
constructions. K. DZIWEREK, P. FERRELL and E. MEJÍAS (eds), Grammar relations. A
cross-theoretical perspective, pp. 109-125. Stanford: The Stanford Linguistic
Association.
DUBINSKY, Stanley. 1990. Light verbs and predicate demotion in Japanese. K.
DZIWEREK, P. FERRELL y E. MEJÍAS (eds), Grammar relations. A crosstheoretical perspective, pp. 127-145. Stanford: The Stanford Linguistic
Association.
DUBINSKY, Stanley. 1994. Syntactic Underspecification : A Minimalist Approach to
Light Verbs. In M. Koizumi, and H. Ura (eds.), Formal Approaches to Japanese
Linguistics, pp. 61-81. Cambridge: MIT Working Papers in Linguistics.
FILLMORE, Charles J. 1968. The case for case. In Bach and Harms (ed.) Universals in
Linguistic Theory, pp. 1-88. New York: Holt, Rinehart, and Winston.
45
Rodrigo Romero-Méndez
FOLEY, William & Robert D. VAN VALIN, Jr. 1984. Functional syntax and universal
grammar. Cambridge: CUP.
GÓMEZ TÓRREGO, Leonardo. 1999. Los verbos auxiliares. Las perífrasis verbales de
infinitivo. In I. Bosque & V. Demonte (eds.) Gramática descriptiva de la lengua
española, vo1. 1, pp. 3323-3390. Madrid: Espasa Calpe.
GRIMSHAW, Jane & Armin MESTER. 1988. “Light verbs and θ–marking”. Linguistic
Inquiry 19 (2), pp. 205-232.
JESPERSEN, Otto. 1964. Essentials of English grammar. Alabama: University of Alabama
Press.
KEMMER, Suzanne. 1993. The middle voice. Amsterdam/Philadelphia: John Benjamins.
KRIFKA, Manfred. 1992. Thematic relations as links between nominal reference and
temporal constitution. In I.Sag and A. Szabolcsi (eds.), Lexical Matters, 29-53.
Stanford, CA: CSLI Publications
LANGACKER, Ronald W. 1987. Nouns and Verbs, Language 63(1), pp. 53-94.
LEVIN, Beth. 1989. Basque verbal inventory and configurationality. In M. Marácz and P.
Muysken (eds.), Configurationality: the typology of assimetries, pp. 39-62.
Dordrecht: Foris.
MAIRAL USÓN, Ricardo & Pamela FABER. 1992. Functional Grammar and lexical
templates. In R. Mairal and M. J. Perez (eds.) New perspectives on argument
structure in Functional Grammar, pp. 39-94. Berlin: Mouton de Gruyter.
MAIRAL USÓN, Ricardo & Pamela FABER. 2005.Decomposing semantic decomposition:
Towards a semantic metalanguage in RRG. In Proceedings of the RRG '05
46
Spanish Light Verb Constructions
Conference. Available online at:
http://linguistics.buffalo.edu/research/rrg/Proceedings%20of%20RRG05.htm
MALDONADO, Ricardo. 1999. A media voz. Mexico: UNAM.
MOHANAN, Tara. 1997. Multidimensionality of representation: NV complex predicates in
Hindi. In A. Alsina, J. Bresnan, and P.Sells (eds.), Complex predicates, pp. 431471. Stanford: CSLI Publications.
NUNES, Mary L. 1993. Argument linking in English Derived Nominals. In R. D. Van
Valin Jr. (ed.), Advances in Role and Reference Grammar, pp. 375-432.
Amsterdam/Philadelphia: John Benjamins.
PARIS, Luis A. 1999. The Spanish causative construction ‘hacer-infinitive’. A Role and
Reference Grammar Description. Unpublished Qualifying Paper, University at
Buffalo.
PARK, Ki-Seon. 1995. The semantics and pragmatics of case marking in Korean: a Role
and Reference Grammar Account. Unpublished Dissertation, University at
Buffalo.
PELLETIER, Rosanne. 1990. Light verbs in Telugu: A clause union analysis. K.
DZIWEREK, P. FERRELL y E. MEJÍAS (eds), Grammar relations. A crosstheoretical perspective, pp. 335-347. Stanford: The Stanford Linguistic
Association.
PENSALFINI, Rob. 2004. Towards a Typology of Configurationality, Natural Language &
Linguistic Theory 22, pp. 359-408.
PICALLO, M. Carme. 1999. La estructura del sintagma nominal: Las nominalizaciones y
otros sustantivos con complementos argumentales. In I. Bosque & V. Demonte
47
Rodrigo Romero-Méndez
(eds.) Gramática descriptiva de la lengua española, vo1. 1, pp. 363-393. Madrid:
RAE-Espasa Calpe.
PICALLO, M. Carme. 1999. La estructura del sintagma nominal: las nominalizaciones y
otros sustantivos con complementos argumentales. I Bosque & V. Demonte (eds),
Gramática descriptiva de la lengua española, Vol. 1, pp. 363-393. Madrid:
Espasa.
PIERA, Carlos & Varela, SOLEDAD. 1999. Relaciones entre morfología y sintaxis. In I.
Bosque & V. Demonte (eds.) Gramática descriptiva de la lengua española, vo1.
1, pp. 4367-4422. Madrid: Espasa Calpe
PORTERO MUÑOZ, Carmen. 2003. Derived nominalizations in -ee: a Role and Reference
Grammar based semantic analysis. English Language and Linguistics 7 (1), pp.
129–159.
PUSTEJOVSKY, James. 1991. The Generative Lexicon, Computational Linguistics 17, pp.
409-441.
PUSTEJOVSKY, James. 1995. Generative Lexicon. Cambrige, Mass: MIT Press.
PUSTEJOVSKY, James. 1998. The Semantics of Lexical Underspecification, Folia
Linguistica 32, 323-347.
Pustejovsky, James. 2000. Syntagmatic Processes. Handbook of Lexicology and
Lexicography.
PUSTEJOVSKY, James. 2001. Type Construction and the Logic of Concepts. In P. Bouillon
and F. Busa (eds.) The language of word meaning, pp. 91-123. Cambridge, UK:
CUP.
48
Spanish Light Verb Constructions
REIHMANN, Sussanne Z. 2001. A constructional approach to idioms and word formation.
Ph.D. dissertation, Stanford University.
ROSEN, Sara Thomas. 1990. Argument structure and Complex predicates. New York:
Garland Publishing.
SAMEK-LODOVICI, Viery. 2003. The Internal Structure of Arguments and its Role in
Complex Predicate Formation, Natural Language and Linguistic Theory 21, 835881.
SMITH, Carlotta. 1997. The Parameter of Aspect. Dordrecht: Reidel.
VAN VALIN, Robert D., Jr. 1993. A synopsis of Role and Reference Grammar. In R.D.
Van Valin Jr. (ed.), Advances in Role and Reference Grammar, pp. 1-164.
Amsterdam/Philadelphia: John Benjamins.
VAN VALIN, Robert D., Jr. 2004. Lexical Representation, Co-composition, and Linking
Syntax and Semantics. Manuscript, University at Buffalo.
(http://linguistics.buffalo.edu/research/rrg/vanvalin_papers/LexRepCoCompLnkg
RRG.pdf)
VAN VALIN, Robert D., Jr. 2005. Exploring the syntax-semantics interface. Cambridge:
CUP.
VAN VALIN, Robert D. & David WILKINS. 1993. Predicting syntactic Structure from
semantic representations: remember in English and its equivalents in Mparntwe
Arrernte. In R.D Van Valin Jr. (ed.), Advances in Role and Reference Grammar,
pp. 499-533. Amsterdam / Philadelphia: John Benjamins.
VAN VALIN, Robert D. Jr & David P. WILKINS. 1996. The Case for ‘Effector’: Case
Roles, Agents and Agency Revisited. M. Shibatani and S.A. Thompson (eds.)
49
Rodrigo Romero-Méndez
Grammatical Constructions: Their Form and Meaning, pp. 289-322. Oxford:
Oxford University Press
VAN VALIN, Robert D., Jr & Randy LAPOLLA. 1997. Syntax. Structure, Meaning, and
Function. Cambridge: Cambridge University Press.
VENDLER, Zeno. 1967. Linguistics in Philosophy. Ithaca: Cornell University Press.
WIERZBICKA, Anna. 1982. Why Can You Have a Drink When You Can’t *Have an Eat?,
Language 58(8), pp.753-799.
50